The relation between identification and discrimination of vowels by young normal-hearing and elderly hearing-impaired listeners

Four young normal-hearing (YNH) and four elderly hearing-impaired subjects (EHI) with moderate sloping sensori-neural hearing losses participated in vowel-identification and formant-discrimination tasks. To examine the relationship between vowel-identification and formant-discrimination abilities in conditions differing in audibility, signals were presented at two levels (70 and 95 dB SPL). Four mid-vowels (/I/, /e/, /ε/, and /æ/) were chosen as the target signals for both tasks. Identification performance for the YNH subjects was near ceiling performance. The EHI subjects averaged 80% for the 95 dB and 69% for the 70 dB SPL presentation level, although individual subject variability was high. Equivalent discrimination performance in the F1 region (ΔF threshold approximately 30 Hz) was observed for all four vowels, between groups and across levels. In the F2 region the EHI subjects’ thresholds were elevated compared to the YNH subjects at both levels, even when the formants appeared to be fully audible (at 95 dB SPL). Correlational analyses suggested that vowel identification was partially predicted by reduced ability to discriminate spectral differences in the F2 region (at higher frequencies) as well as the subjects’ hearing loss. [Work supported by NIH and NIA.]